The Science Council (n.d.), a body overseeing scientific standards and practice within the UK, came up with one definition:
Science is the pursuit of knowledge and understanding of the natural and social world following a systematic methodology based on evidence.
Hofman et al. (2021)’s 2x2:
| No intervention or distributional changes | Under interventions or distributional changes | |
|---|---|---|
| Focus on specific features or effects | Descriptive modelling (1) | Explanatory modelling (2) |
| Focus on predicting outcomes | Predictive modelling (3) | Integrative modelling (4) |
| Granularity | Descriptive modelling | Explanatory modelling | Predictive modelling | Integrative modelling |
|---|---|---|---|---|
| Describes something | Tests a causal claim | Tests a (passive) predictive claim | Tests a claim both for causality and predictive accuracy | |
| Low | Reports stylized facts | Tests for a non-zero effect | Predicts directional or aggregate outcomes | Predicts directional or aggregate outcomes under changes or interventions |
| Medium | Reports population averages | Tests for a directional effect | Predicts magnitude and direction of aggregate outcomes | Predicts magnitude and direction of aggregate outcomes under changes or interventions |
| High | Reports individual outcomes | Estimates the magnitude and direction of an effect | Predicts magnitude and direction of individual outcomes | Predicts magnitude and direction of individual outcomes under changes or interventions |